I just wrote a piece similar to this last week, but this one drives the point home even more. Basically, the current trajectory of AI, with all the art generation, the language models, etc., are about to become a whole lot more instruction and response based. What does that mean?
It means rather than having to trick your AI Art generator into making something cool using special prompt wizardry, you’re going to be able to give a similar model basic instructions to do pretty much anything. Write the perfect email for my boss. Find the best ideas in the company. Tell me what customers are closest to churning and what would win them back. Etc. It will be able to do this because it can break down a bigger task into smaller, required tasks that accomplish it. And it will simply produce magic.
We can now make agents that, on the basis of natural language questions or instructions, can perform actions in a wide action space. The potential is not limited to CX or sales any particular business workflow. Any set of behaviors where an agent manipulates symbols in a computer can in principle be performed with this schema. And it looks an awful lot like AGI.
John McDonnell
I’ve been doing this for months already using GPT-3, and I’m completely stunned by what it can do with, say, a security news story. I can give it the body of an article and it can tell me who the attacker was, who the defender was, what technique they used in their attack, and tons of other important analysis. It’s insane. And to the point of my previous article, it’s precisely what we thought could only come from an AGI.
But guess what? Most people are about to get a whole lot less interested in AGI, because we’re about to get the benefits of AGI without it needing to be self-aware. AGI-level powers are simply emerging from these evolutions in transformers and LLMs.
I’m telling you again, and please listen. I know how hyperbolic this sounds. It’s not hyperbole. This is bigger than the internet. This is about to change everything because it is changing the fundamental value of human work. We’re talking about the ability to do pretty much any knowledge work task better than humans.
You need to get ready. We all do.